xref: /llvm-project/llvm/test/CodeGen/AArch64/arm64-memset-inline.ll (revision 6e54fccede402c9ed0e8038aa258a99c5a2773e5) 
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2; RUN: llc < %s -mtriple=arm64-eabi | FileCheck %s
3
4define void @bzero_4_heap(ptr nocapture %c) {
5; CHECK-LABEL: bzero_4_heap:
6; CHECK:       // %bb.0:
7; CHECK-NEXT:    str wzr, [x0]
8; CHECK-NEXT:    ret
9  call void @llvm.memset.p0.i64(ptr align 4 %c, i8 0, i64 4, i1 false)
10  ret void
11}
12
13define void @bzero_8_heap(ptr nocapture %c) {
14; CHECK-LABEL: bzero_8_heap:
15; CHECK:       // %bb.0:
16; CHECK-NEXT:    str xzr, [x0]
17; CHECK-NEXT:    ret
18  call void @llvm.memset.p0.i64(ptr align 8 %c, i8 0, i64 8, i1 false)
19  ret void
20}
21
22define void @bzero_12_heap(ptr nocapture %c) {
23; CHECK-LABEL: bzero_12_heap:
24; CHECK:       // %bb.0:
25; CHECK-NEXT:    str wzr, [x0, #8]
26; CHECK-NEXT:    str xzr, [x0]
27; CHECK-NEXT:    ret
28  call void @llvm.memset.p0.i64(ptr align 8 %c, i8 0, i64 12, i1 false)
29  ret void
30}
31
32define void @bzero_16_heap(ptr nocapture %c) {
33; CHECK-LABEL: bzero_16_heap:
34; CHECK:       // %bb.0:
35; CHECK-NEXT:    stp xzr, xzr, [x0]
36; CHECK-NEXT:    ret
37  call void @llvm.memset.p0.i64(ptr align 8 %c, i8 0, i64 16, i1 false)
38  ret void
39}
40
41define void @bzero_32_heap(ptr nocapture %c) {
42; CHECK-LABEL: bzero_32_heap:
43; CHECK:       // %bb.0:
44; CHECK-NEXT:    movi v0.2d, #0000000000000000
45; CHECK-NEXT:    stp q0, q0, [x0]
46; CHECK-NEXT:    ret
47  call void @llvm.memset.p0.i64(ptr align 8 %c, i8 0, i64 32, i1 false)
48  ret void
49}
50
51define void @bzero_64_heap(ptr nocapture %c) {
52; CHECK-LABEL: bzero_64_heap:
53; CHECK:       // %bb.0:
54; CHECK-NEXT:    movi v0.2d, #0000000000000000
55; CHECK-NEXT:    stp q0, q0, [x0]
56; CHECK-NEXT:    stp q0, q0, [x0, #32]
57; CHECK-NEXT:    ret
58  call void @llvm.memset.p0.i64(ptr align 8 %c, i8 0, i64 64, i1 false)
59  ret void
60}
61
62define void @bzero_4_stack() {
63; CHECK-LABEL: bzero_4_stack:
64; CHECK:       // %bb.0:
65; CHECK-NEXT:    str x30, [sp, #-16]! // 8-byte Folded Spill
66; CHECK-NEXT:    .cfi_def_cfa_offset 16
67; CHECK-NEXT:    .cfi_offset w30, -16
68; CHECK-NEXT:    add x0, sp, #12
69; CHECK-NEXT:    str wzr, [sp, #12]
70; CHECK-NEXT:    bl something
71; CHECK-NEXT:    ldr x30, [sp], #16 // 8-byte Folded Reload
72; CHECK-NEXT:    ret
73  %buf = alloca [4 x i8], align 1
74  call void @llvm.memset.p0.i32(ptr %buf, i8 0, i32 4, i1 false)
75  call void @something(ptr %buf)
76  ret void
77}
78
79define void @bzero_8_stack() {
80; CHECK-LABEL: bzero_8_stack:
81; CHECK:       // %bb.0:
82; CHECK-NEXT:    stp x30, xzr, [sp, #-16]! // 8-byte Folded Spill
83; CHECK-NEXT:    .cfi_def_cfa_offset 16
84; CHECK-NEXT:    .cfi_offset w30, -16
85; CHECK-NEXT:    add x0, sp, #8
86; CHECK-NEXT:    bl something
87; CHECK-NEXT:    ldr x30, [sp], #16 // 8-byte Folded Reload
88; CHECK-NEXT:    ret
89  %buf = alloca [8 x i8], align 1
90  call void @llvm.memset.p0.i32(ptr %buf, i8 0, i32 8, i1 false)
91  call void @something(ptr %buf)
92  ret void
93}
94
95define void @bzero_12_stack() {
96; CHECK-LABEL: bzero_12_stack:
97; CHECK:       // %bb.0:
98; CHECK-NEXT:    sub sp, sp, #32
99; CHECK-NEXT:    str x30, [sp, #16] // 8-byte Folded Spill
100; CHECK-NEXT:    .cfi_def_cfa_offset 32
101; CHECK-NEXT:    .cfi_offset w30, -16
102; CHECK-NEXT:    mov x0, sp
103; CHECK-NEXT:    str wzr, [sp, #8]
104; CHECK-NEXT:    str xzr, [sp]
105; CHECK-NEXT:    bl something
106; CHECK-NEXT:    ldr x30, [sp, #16] // 8-byte Folded Reload
107; CHECK-NEXT:    add sp, sp, #32
108; CHECK-NEXT:    ret
109  %buf = alloca [12 x i8], align 1
110  call void @llvm.memset.p0.i32(ptr %buf, i8 0, i32 12, i1 false)
111  call void @something(ptr %buf)
112  ret void
113}
114
115define void @bzero_16_stack() {
116; CHECK-LABEL: bzero_16_stack:
117; CHECK:       // %bb.0:
118; CHECK-NEXT:    sub sp, sp, #32
119; CHECK-NEXT:    stp xzr, x30, [sp, #8] // 8-byte Folded Spill
120; CHECK-NEXT:    .cfi_def_cfa_offset 32
121; CHECK-NEXT:    .cfi_offset w30, -16
122; CHECK-NEXT:    mov x0, sp
123; CHECK-NEXT:    str xzr, [sp]
124; CHECK-NEXT:    bl something
125; CHECK-NEXT:    ldr x30, [sp, #16] // 8-byte Folded Reload
126; CHECK-NEXT:    add sp, sp, #32
127; CHECK-NEXT:    ret
128  %buf = alloca [16 x i8], align 1
129  call void @llvm.memset.p0.i32(ptr %buf, i8 0, i32 16, i1 false)
130  call void @something(ptr %buf)
131  ret void
132}
133
134define void @bzero_20_stack() {
135; CHECK-LABEL: bzero_20_stack:
136; CHECK:       // %bb.0:
137; CHECK-NEXT:    sub sp, sp, #48
138; CHECK-NEXT:    str x30, [sp, #32] // 8-byte Folded Spill
139; CHECK-NEXT:    .cfi_def_cfa_offset 48
140; CHECK-NEXT:    .cfi_offset w30, -16
141; CHECK-NEXT:    add x0, sp, #8
142; CHECK-NEXT:    stp xzr, xzr, [sp, #8]
143; CHECK-NEXT:    str wzr, [sp, #24]
144; CHECK-NEXT:    bl something
145; CHECK-NEXT:    ldr x30, [sp, #32] // 8-byte Folded Reload
146; CHECK-NEXT:    add sp, sp, #48
147; CHECK-NEXT:    ret
148  %buf = alloca [20 x i8], align 1
149  call void @llvm.memset.p0.i32(ptr %buf, i8 0, i32 20, i1 false)
150  call void @something(ptr %buf)
151  ret void
152}
153
154define void @bzero_26_stack() {
155; CHECK-LABEL: bzero_26_stack:
156; CHECK:       // %bb.0:
157; CHECK-NEXT:    sub sp, sp, #48
158; CHECK-NEXT:    str x30, [sp, #32] // 8-byte Folded Spill
159; CHECK-NEXT:    .cfi_def_cfa_offset 48
160; CHECK-NEXT:    .cfi_offset w30, -16
161; CHECK-NEXT:    mov x0, sp
162; CHECK-NEXT:    stp xzr, xzr, [sp]
163; CHECK-NEXT:    strh wzr, [sp, #24]
164; CHECK-NEXT:    str xzr, [sp, #16]
165; CHECK-NEXT:    bl something
166; CHECK-NEXT:    ldr x30, [sp, #32] // 8-byte Folded Reload
167; CHECK-NEXT:    add sp, sp, #48
168; CHECK-NEXT:    ret
169  %buf = alloca [26 x i8], align 1
170  call void @llvm.memset.p0.i32(ptr %buf, i8 0, i32 26, i1 false)
171  call void @something(ptr %buf)
172  ret void
173}
174
175define void @bzero_32_stack() {
176; CHECK-LABEL: bzero_32_stack:
177; CHECK:       // %bb.0:
178; CHECK-NEXT:    sub sp, sp, #48
179; CHECK-NEXT:    str x30, [sp, #32] // 8-byte Folded Spill
180; CHECK-NEXT:    .cfi_def_cfa_offset 48
181; CHECK-NEXT:    .cfi_offset w30, -16
182; CHECK-NEXT:    movi v0.2d, #0000000000000000
183; CHECK-NEXT:    mov x0, sp
184; CHECK-NEXT:    stp q0, q0, [sp]
185; CHECK-NEXT:    bl something
186; CHECK-NEXT:    ldr x30, [sp, #32] // 8-byte Folded Reload
187; CHECK-NEXT:    add sp, sp, #48
188; CHECK-NEXT:    ret
189  %buf = alloca [32 x i8], align 1
190  call void @llvm.memset.p0.i32(ptr %buf, i8 0, i32 32, i1 false)
191  call void @something(ptr %buf)
192  ret void
193}
194
195define void @bzero_40_stack() {
196; CHECK-LABEL: bzero_40_stack:
197; CHECK:       // %bb.0:
198; CHECK-NEXT:    sub sp, sp, #64
199; CHECK-NEXT:    str x30, [sp, #48] // 8-byte Folded Spill
200; CHECK-NEXT:    .cfi_def_cfa_offset 64
201; CHECK-NEXT:    .cfi_offset w30, -16
202; CHECK-NEXT:    movi v0.2d, #0000000000000000
203; CHECK-NEXT:    mov x0, sp
204; CHECK-NEXT:    str xzr, [sp, #32]
205; CHECK-NEXT:    stp q0, q0, [sp]
206; CHECK-NEXT:    bl something
207; CHECK-NEXT:    ldr x30, [sp, #48] // 8-byte Folded Reload
208; CHECK-NEXT:    add sp, sp, #64
209; CHECK-NEXT:    ret
210  %buf = alloca [40 x i8], align 1
211  call void @llvm.memset.p0.i32(ptr %buf, i8 0, i32 40, i1 false)
212  call void @something(ptr %buf)
213  ret void
214}
215
216define void @bzero_64_stack() {
217; CHECK-LABEL: bzero_64_stack:
218; CHECK:       // %bb.0:
219; CHECK-NEXT:    sub sp, sp, #80
220; CHECK-NEXT:    str x30, [sp, #64] // 8-byte Folded Spill
221; CHECK-NEXT:    .cfi_def_cfa_offset 80
222; CHECK-NEXT:    .cfi_offset w30, -16
223; CHECK-NEXT:    movi v0.2d, #0000000000000000
224; CHECK-NEXT:    mov x0, sp
225; CHECK-NEXT:    stp q0, q0, [sp]
226; CHECK-NEXT:    stp q0, q0, [sp, #32]
227; CHECK-NEXT:    bl something
228; CHECK-NEXT:    ldr x30, [sp, #64] // 8-byte Folded Reload
229; CHECK-NEXT:    add sp, sp, #80
230; CHECK-NEXT:    ret
231  %buf = alloca [64 x i8], align 1
232  call void @llvm.memset.p0.i32(ptr %buf, i8 0, i32 64, i1 false)
233  call void @something(ptr %buf)
234  ret void
235}
236
237define void @bzero_72_stack() {
238; CHECK-LABEL: bzero_72_stack:
239; CHECK:       // %bb.0:
240; CHECK-NEXT:    sub sp, sp, #96
241; CHECK-NEXT:    str x30, [sp, #80] // 8-byte Folded Spill
242; CHECK-NEXT:    .cfi_def_cfa_offset 96
243; CHECK-NEXT:    .cfi_offset w30, -16
244; CHECK-NEXT:    movi v0.2d, #0000000000000000
245; CHECK-NEXT:    mov x0, sp
246; CHECK-NEXT:    str xzr, [sp, #64]
247; CHECK-NEXT:    stp q0, q0, [sp]
248; CHECK-NEXT:    stp q0, q0, [sp, #32]
249; CHECK-NEXT:    bl something
250; CHECK-NEXT:    ldr x30, [sp, #80] // 8-byte Folded Reload
251; CHECK-NEXT:    add sp, sp, #96
252; CHECK-NEXT:    ret
253  %buf = alloca [72 x i8], align 1
254  call void @llvm.memset.p0.i32(ptr %buf, i8 0, i32 72, i1 false)
255  call void @something(ptr %buf)
256  ret void
257}
258
259define void @bzero_128_stack() {
260; CHECK-LABEL: bzero_128_stack:
261; CHECK:       // %bb.0:
262; CHECK-NEXT:    sub sp, sp, #144
263; CHECK-NEXT:    str x30, [sp, #128] // 8-byte Folded Spill
264; CHECK-NEXT:    .cfi_def_cfa_offset 144
265; CHECK-NEXT:    .cfi_offset w30, -16
266; CHECK-NEXT:    movi v0.2d, #0000000000000000
267; CHECK-NEXT:    mov x0, sp
268; CHECK-NEXT:    stp q0, q0, [sp]
269; CHECK-NEXT:    stp q0, q0, [sp, #32]
270; CHECK-NEXT:    stp q0, q0, [sp, #64]
271; CHECK-NEXT:    stp q0, q0, [sp, #96]
272; CHECK-NEXT:    bl something
273; CHECK-NEXT:    ldr x30, [sp, #128] // 8-byte Folded Reload
274; CHECK-NEXT:    add sp, sp, #144
275; CHECK-NEXT:    ret
276  %buf = alloca [128 x i8], align 1
277  call void @llvm.memset.p0.i32(ptr %buf, i8 0, i32 128, i1 false)
278  call void @something(ptr %buf)
279  ret void
280}
281
282define void @bzero_256_stack() {
283; CHECK-LABEL: bzero_256_stack:
284; CHECK:       // %bb.0:
285; CHECK-NEXT:    sub sp, sp, #272
286; CHECK-NEXT:    stp x29, x30, [sp, #256] // 16-byte Folded Spill
287; CHECK-NEXT:    .cfi_def_cfa_offset 272
288; CHECK-NEXT:    .cfi_offset w30, -8
289; CHECK-NEXT:    .cfi_offset w29, -16
290; CHECK-NEXT:    movi v0.2d, #0000000000000000
291; CHECK-NEXT:    mov x0, sp
292; CHECK-NEXT:    stp q0, q0, [sp]
293; CHECK-NEXT:    stp q0, q0, [sp, #32]
294; CHECK-NEXT:    stp q0, q0, [sp, #64]
295; CHECK-NEXT:    stp q0, q0, [sp, #96]
296; CHECK-NEXT:    stp q0, q0, [sp, #128]
297; CHECK-NEXT:    stp q0, q0, [sp, #160]
298; CHECK-NEXT:    stp q0, q0, [sp, #192]
299; CHECK-NEXT:    stp q0, q0, [sp, #224]
300; CHECK-NEXT:    bl something
301; CHECK-NEXT:    ldp x29, x30, [sp, #256] // 16-byte Folded Reload
302; CHECK-NEXT:    add sp, sp, #272
303; CHECK-NEXT:    ret
304  %buf = alloca [256 x i8], align 1
305  call void @llvm.memset.p0.i32(ptr %buf, i8 0, i32 256, i1 false)
306  call void @something(ptr %buf)
307  ret void
308}
309
310define void @memset_4_stack() {
311; CHECK-LABEL: memset_4_stack:
312; CHECK:       // %bb.0:
313; CHECK-NEXT:    str x30, [sp, #-16]! // 8-byte Folded Spill
314; CHECK-NEXT:    .cfi_def_cfa_offset 16
315; CHECK-NEXT:    .cfi_offset w30, -16
316; CHECK-NEXT:    mov w8, #-1431655766
317; CHECK-NEXT:    add x0, sp, #12
318; CHECK-NEXT:    str w8, [sp, #12]
319; CHECK-NEXT:    bl something
320; CHECK-NEXT:    ldr x30, [sp], #16 // 8-byte Folded Reload
321; CHECK-NEXT:    ret
322  %buf = alloca [4 x i8], align 1
323  call void @llvm.memset.p0.i32(ptr %buf, i8 -86, i32 4, i1 false)
324  call void @something(ptr %buf)
325  ret void
326}
327
328define void @memset_8_stack() {
329; CHECK-LABEL: memset_8_stack:
330; CHECK:       // %bb.0:
331; CHECK-NEXT:    .cfi_def_cfa_offset 16
332; CHECK-NEXT:    .cfi_offset w30, -16
333; CHECK-NEXT:    mov x8, #-6148914691236517206
334; CHECK-NEXT:    stp x30, x8, [sp, #-16]! // 8-byte Folded Spill
335; CHECK-NEXT:    add x0, sp, #8
336; CHECK-NEXT:    bl something
337; CHECK-NEXT:    ldr x30, [sp], #16 // 8-byte Folded Reload
338; CHECK-NEXT:    ret
339  %buf = alloca [8 x i8], align 1
340  call void @llvm.memset.p0.i32(ptr %buf, i8 -86, i32 8, i1 false)
341  call void @something(ptr %buf)
342  ret void
343}
344
345define void @memset_12_stack() {
346; CHECK-LABEL: memset_12_stack:
347; CHECK:       // %bb.0:
348; CHECK-NEXT:    sub sp, sp, #32
349; CHECK-NEXT:    str x30, [sp, #16] // 8-byte Folded Spill
350; CHECK-NEXT:    .cfi_def_cfa_offset 32
351; CHECK-NEXT:    .cfi_offset w30, -16
352; CHECK-NEXT:    mov x8, #-6148914691236517206
353; CHECK-NEXT:    mov x0, sp
354; CHECK-NEXT:    str x8, [sp]
355; CHECK-NEXT:    str w8, [sp, #8]
356; CHECK-NEXT:    bl something
357; CHECK-NEXT:    ldr x30, [sp, #16] // 8-byte Folded Reload
358; CHECK-NEXT:    add sp, sp, #32
359; CHECK-NEXT:    ret
360  %buf = alloca [12 x i8], align 1
361  call void @llvm.memset.p0.i32(ptr %buf, i8 -86, i32 12, i1 false)
362  call void @something(ptr %buf)
363  ret void
364}
365
366define void @memset_16_stack() {
367; CHECK-LABEL: memset_16_stack:
368; CHECK:       // %bb.0:
369; CHECK-NEXT:    sub sp, sp, #32
370; CHECK-NEXT:    .cfi_def_cfa_offset 32
371; CHECK-NEXT:    .cfi_offset w30, -16
372; CHECK-NEXT:    mov x8, #-6148914691236517206
373; CHECK-NEXT:    mov x0, sp
374; CHECK-NEXT:    stp x8, x30, [sp, #8] // 8-byte Folded Spill
375; CHECK-NEXT:    str x8, [sp]
376; CHECK-NEXT:    bl something
377; CHECK-NEXT:    ldr x30, [sp, #16] // 8-byte Folded Reload
378; CHECK-NEXT:    add sp, sp, #32
379; CHECK-NEXT:    ret
380  %buf = alloca [16 x i8], align 1
381  call void @llvm.memset.p0.i32(ptr %buf, i8 -86, i32 16, i1 false)
382  call void @something(ptr %buf)
383  ret void
384}
385
386define void @memset_20_stack() {
387; CHECK-LABEL: memset_20_stack:
388; CHECK:       // %bb.0:
389; CHECK-NEXT:    sub sp, sp, #48
390; CHECK-NEXT:    str x30, [sp, #32] // 8-byte Folded Spill
391; CHECK-NEXT:    .cfi_def_cfa_offset 48
392; CHECK-NEXT:    .cfi_offset w30, -16
393; CHECK-NEXT:    mov x8, #-6148914691236517206
394; CHECK-NEXT:    add x0, sp, #8
395; CHECK-NEXT:    stp x8, x8, [sp, #8]
396; CHECK-NEXT:    str w8, [sp, #24]
397; CHECK-NEXT:    bl something
398; CHECK-NEXT:    ldr x30, [sp, #32] // 8-byte Folded Reload
399; CHECK-NEXT:    add sp, sp, #48
400; CHECK-NEXT:    ret
401  %buf = alloca [20 x i8], align 1
402  call void @llvm.memset.p0.i32(ptr %buf, i8 -86, i32 20, i1 false)
403  call void @something(ptr %buf)
404  ret void
405}
406
407define void @memset_26_stack() {
408; CHECK-LABEL: memset_26_stack:
409; CHECK:       // %bb.0:
410; CHECK-NEXT:    sub sp, sp, #48
411; CHECK-NEXT:    str x30, [sp, #32] // 8-byte Folded Spill
412; CHECK-NEXT:    .cfi_def_cfa_offset 48
413; CHECK-NEXT:    .cfi_offset w30, -16
414; CHECK-NEXT:    mov x8, #-6148914691236517206
415; CHECK-NEXT:    mov x0, sp
416; CHECK-NEXT:    stp x8, x8, [sp, #8]
417; CHECK-NEXT:    str x8, [sp]
418; CHECK-NEXT:    strh w8, [sp, #24]
419; CHECK-NEXT:    bl something
420; CHECK-NEXT:    ldr x30, [sp, #32] // 8-byte Folded Reload
421; CHECK-NEXT:    add sp, sp, #48
422; CHECK-NEXT:    ret
423  %buf = alloca [26 x i8], align 1
424  call void @llvm.memset.p0.i32(ptr %buf, i8 -86, i32 26, i1 false)
425  call void @something(ptr %buf)
426  ret void
427}
428
429define void @memset_32_stack() {
430; CHECK-LABEL: memset_32_stack:
431; CHECK:       // %bb.0:
432; CHECK-NEXT:    sub sp, sp, #48
433; CHECK-NEXT:    str x30, [sp, #32] // 8-byte Folded Spill
434; CHECK-NEXT:    .cfi_def_cfa_offset 48
435; CHECK-NEXT:    .cfi_offset w30, -16
436; CHECK-NEXT:    movi v0.16b, #170
437; CHECK-NEXT:    mov x0, sp
438; CHECK-NEXT:    stp q0, q0, [sp]
439; CHECK-NEXT:    bl something
440; CHECK-NEXT:    ldr x30, [sp, #32] // 8-byte Folded Reload
441; CHECK-NEXT:    add sp, sp, #48
442; CHECK-NEXT:    ret
443  %buf = alloca [32 x i8], align 1
444  call void @llvm.memset.p0.i32(ptr %buf, i8 -86, i32 32, i1 false)
445  call void @something(ptr %buf)
446  ret void
447}
448
449define void @memset_40_stack() {
450; CHECK-LABEL: memset_40_stack:
451; CHECK:       // %bb.0:
452; CHECK-NEXT:    sub sp, sp, #64
453; CHECK-NEXT:    str x30, [sp, #48] // 8-byte Folded Spill
454; CHECK-NEXT:    .cfi_def_cfa_offset 64
455; CHECK-NEXT:    .cfi_offset w30, -16
456; CHECK-NEXT:    movi v0.16b, #170
457; CHECK-NEXT:    mov x8, #-6148914691236517206
458; CHECK-NEXT:    mov x0, sp
459; CHECK-NEXT:    str x8, [sp, #32]
460; CHECK-NEXT:    stp q0, q0, [sp]
461; CHECK-NEXT:    bl something
462; CHECK-NEXT:    ldr x30, [sp, #48] // 8-byte Folded Reload
463; CHECK-NEXT:    add sp, sp, #64
464; CHECK-NEXT:    ret
465  %buf = alloca [40 x i8], align 1
466  call void @llvm.memset.p0.i32(ptr %buf, i8 -86, i32 40, i1 false)
467  call void @something(ptr %buf)
468  ret void
469}
470
471define void @memset_64_stack() {
472; CHECK-LABEL: memset_64_stack:
473; CHECK:       // %bb.0:
474; CHECK-NEXT:    sub sp, sp, #80
475; CHECK-NEXT:    str x30, [sp, #64] // 8-byte Folded Spill
476; CHECK-NEXT:    .cfi_def_cfa_offset 80
477; CHECK-NEXT:    .cfi_offset w30, -16
478; CHECK-NEXT:    movi v0.16b, #170
479; CHECK-NEXT:    mov x0, sp
480; CHECK-NEXT:    stp q0, q0, [sp]
481; CHECK-NEXT:    stp q0, q0, [sp, #32]
482; CHECK-NEXT:    bl something
483; CHECK-NEXT:    ldr x30, [sp, #64] // 8-byte Folded Reload
484; CHECK-NEXT:    add sp, sp, #80
485; CHECK-NEXT:    ret
486  %buf = alloca [64 x i8], align 1
487  call void @llvm.memset.p0.i32(ptr %buf, i8 -86, i32 64, i1 false)
488  call void @something(ptr %buf)
489  ret void
490}
491
492define void @memset_72_stack() {
493; CHECK-LABEL: memset_72_stack:
494; CHECK:       // %bb.0:
495; CHECK-NEXT:    sub sp, sp, #96
496; CHECK-NEXT:    str x30, [sp, #80] // 8-byte Folded Spill
497; CHECK-NEXT:    .cfi_def_cfa_offset 96
498; CHECK-NEXT:    .cfi_offset w30, -16
499; CHECK-NEXT:    movi v0.16b, #170
500; CHECK-NEXT:    mov x8, #-6148914691236517206
501; CHECK-NEXT:    mov x0, sp
502; CHECK-NEXT:    str x8, [sp, #64]
503; CHECK-NEXT:    stp q0, q0, [sp]
504; CHECK-NEXT:    stp q0, q0, [sp, #32]
505; CHECK-NEXT:    bl something
506; CHECK-NEXT:    ldr x30, [sp, #80] // 8-byte Folded Reload
507; CHECK-NEXT:    add sp, sp, #96
508; CHECK-NEXT:    ret
509  %buf = alloca [72 x i8], align 1
510  call void @llvm.memset.p0.i32(ptr %buf, i8 -86, i32 72, i1 false)
511  call void @something(ptr %buf)
512  ret void
513}
514
515define void @memset_128_stack() {
516; CHECK-LABEL: memset_128_stack:
517; CHECK:       // %bb.0:
518; CHECK-NEXT:    sub sp, sp, #144
519; CHECK-NEXT:    str x30, [sp, #128] // 8-byte Folded Spill
520; CHECK-NEXT:    .cfi_def_cfa_offset 144
521; CHECK-NEXT:    .cfi_offset w30, -16
522; CHECK-NEXT:    movi v0.16b, #170
523; CHECK-NEXT:    mov x0, sp
524; CHECK-NEXT:    stp q0, q0, [sp]
525; CHECK-NEXT:    stp q0, q0, [sp, #32]
526; CHECK-NEXT:    stp q0, q0, [sp, #64]
527; CHECK-NEXT:    stp q0, q0, [sp, #96]
528; CHECK-NEXT:    bl something
529; CHECK-NEXT:    ldr x30, [sp, #128] // 8-byte Folded Reload
530; CHECK-NEXT:    add sp, sp, #144
531; CHECK-NEXT:    ret
532  %buf = alloca [128 x i8], align 1
533  call void @llvm.memset.p0.i32(ptr %buf, i8 -86, i32 128, i1 false)
534  call void @something(ptr %buf)
535  ret void
536}
537
538define void @memset_256_stack() {
539; CHECK-LABEL: memset_256_stack:
540; CHECK:       // %bb.0:
541; CHECK-NEXT:    sub sp, sp, #272
542; CHECK-NEXT:    stp x29, x30, [sp, #256] // 16-byte Folded Spill
543; CHECK-NEXT:    .cfi_def_cfa_offset 272
544; CHECK-NEXT:    .cfi_offset w30, -8
545; CHECK-NEXT:    .cfi_offset w29, -16
546; CHECK-NEXT:    movi v0.16b, #170
547; CHECK-NEXT:    mov x0, sp
548; CHECK-NEXT:    stp q0, q0, [sp]
549; CHECK-NEXT:    stp q0, q0, [sp, #32]
550; CHECK-NEXT:    stp q0, q0, [sp, #64]
551; CHECK-NEXT:    stp q0, q0, [sp, #96]
552; CHECK-NEXT:    stp q0, q0, [sp, #128]
553; CHECK-NEXT:    stp q0, q0, [sp, #160]
554; CHECK-NEXT:    stp q0, q0, [sp, #192]
555; CHECK-NEXT:    stp q0, q0, [sp, #224]
556; CHECK-NEXT:    bl something
557; CHECK-NEXT:    ldp x29, x30, [sp, #256] // 16-byte Folded Reload
558; CHECK-NEXT:    add sp, sp, #272
559; CHECK-NEXT:    ret
560  %buf = alloca [256 x i8], align 1
561  call void @llvm.memset.p0.i32(ptr %buf, i8 -86, i32 256, i1 false)
562  call void @something(ptr %buf)
563  ret void
564}
565
566declare void @something(ptr)
567declare void @llvm.memset.p0.i32(ptr nocapture, i8, i32, i1) nounwind
568declare void @llvm.memset.p0.i64(ptr nocapture, i8, i64, i1) nounwind
569